The long-term relationship between poverty and debt

James Mitchell, Kostas Mouratidis and Martin Weale

This study analyses the benefits for people on low incomes of using credit to smooth out fluctuations in their incomes.



Credit markets exist to allow people to smooth out the effects of temporary fluctuations in their incomes and it is desirable that people should use them for this purpose. This research investigated how far they do so by looking at changes to reported debts and wealth holdings between 1995 and 2000 as reported in the British Household Panel Survey. The research, by a team from the National Institute of Economic and Social Research, found:

  • Analysis of the data suggested that people in poverty in 2000 tended to have debts relative to their incomes 20 to 25 per cent higher than those of the population as a whole.
  • Borrowing by poor people can be justified by the fact that for most households of working age poverty is only temporary. 35 per cent of households in the sample were in poverty in at least one year between 1991 and 2002 but only 2.5 per cent of households were in poverty for ten years or more.
  • An analysis of how much poor households might have borrowed if they had been able to respond to reasonable forecasts of future income and borrow on reasonable terms suggests that, between 1995 and 2000, households whose incomes rose sharply borrowed considerably less than they might.
  • This suggests that a key issue in alleviating the effects of poverty is increasing access to credit on reasonable terms. Market rates at which high-income people can borrow are well below those offered by the credit unions which the government sees as a means of providing affordable credit to people on low incomes.
  • The researchers suggest that one way for the government to do this would be to explore a substantial expansion of the Social Fund as a means of improving access to credit by people in poverty.


There is considerable public concern about rising levels of household debt. However, much of the discussion assumes that there is something inherently undesirable about borrowing and neglects two good reasons for borrowing: to access higher income expected in the future and to smooth out fluctuations in expenditure.

People whose incomes vary are likely to wish to avoid large fluctuations in their expenditure from one year to the next. It follows that in periods when income is believed to be temporarily low, it may be sensible to run down savings or, for households with no savings, to go into debt. Nevertheless, in most cases people do not know their future incomes with certainty; even if people on low incomes have a reasonable expectation that better times will come, they will not materialise for everyone. Hence, even ‘sensible’ use of credit is bound to lead to a situation where a ‘run of bad luck’ leaves people with debts that are difficult to handle. So it is not indebtedness per se, but rather the pattern of borrowing at different stages in life that can help tell us whether people are incurring debt on a reasonable basis.

This study explores how far changes in household assets and liabilities can be related to movements in household incomes and, in particular, how far people on low incomes borrow relative to a benchmark that can be justified by experience of improving incomes.

The study focuses on households headed by people under the age of 65 and uses the British Household Panel Survey. All income and wealth variables are quoted for a standard household consisting of two adults. Figures for households differently constituted are scaled to this basis using standard methods of adjusting for household size.

As well as mortgages, ‘debt’ was defined as including loans secured on housing and unsecured debt such as hire purchase credit, credit card debt, personal loans, Social Fund loans and mail order debt. ‘Poverty' was defined as having an income below 60 per cent of median income (after adjusting for household size).

Wealth and debt levels

The very uneven distribution of wealth results from specific factors as well as the overall inequalities that exist within society. Three particular factors promote wealth inequality:

  • People save as they age, so that old people have considerably more wealth than young people.
  • The state benefit system is more important for poor people than for high earners, so people on low incomes have less incentive to save. Additionally, means-testing of state benefits helps widen wealth inequalities.
  • Home-ownership is a very important determinant of overall wealth levels and much more so than simply whether households have low incomes in a particular year; Figure 1 shows this. This pattern was observed at all age groups in the survey and does not simply reflect the fact that old people are more likely to own houses than young people. Income poverty and asset poverty are very different things.

Figure 1: Home ownership, poverty and wealth in 2000

Young people in the survey tended to have higher levels of mortgage and non-mortgage debt, relative to their incomes, than older people. People in poverty tended to have levels of overall debt, relative to their incomes, 20 to 25 per cent higher than those of the population as a whole.

Fewer than half of the households in the sample reported unsecured debt. These households had, on average, incomes slightly above those of the rest of the population. However, they had considerably larger debts and lower levels of overall wealth. The differences with the rest of the population became more pronounced with age.

Prudent spending and borrowing

If people want their spending to evolve smoothly and are able to form expectations of their future income and life-span, then it is possible to work out how much they should plan to spend in each period, so that expected income finances planned expenditure. Using the data on how the incomes of individual households actually change over time, based on the twelve years of data in the British Household Panel Survey, we can work out how households in particular circumstances might expect their incomes to change as they age, and so calculate affordable levels of household spending for the households present in the survey.

A key point that emerges is that, between 1995 and 2000, households headed by people of working age on low incomes experienced faster than average rises in their incomes. This is illustrated by Figure 2, which shows median and first quartile incomes of those in the sample. It also shows the median and first quartile of the incomes of those below the 1995 poverty threshold.

  • The income of the median household in the whole sample rose by 17 per cent between 1995 and 2002.
  • But the income of the median household in poverty rose by 67 per cent and the income of the bottom quartile household in poverty rose by 60 per cent over the same period.

Figure 2: Household incomes: 1995-2002, inflation-corrected

It follows that the median household in poverty in 1995 could, provided it had access to credit on reasonable terms, have borrowed a substantial proportion of its (then low) income and managed its debts given its subsequent rise in income. In other words, the levels of consumption that many households in poverty in 1995 could afford were considerably higher than their incomes in the same year.

Given this assessment of how people might expect their incomes to change over time, we are able to calculate the rate at which people should accumulate wealth by saving or reduce it by spending assets or borrowing. This simple analysis makes the assumption that people do not want to leave bequests for their children and that they therefore have the whole of their wealth (including the value of their house) available to support themselves in retirement.

Comparing this calculated value for the change in wealth with what was actually observed reveals a number of important differences. Those who already had greater assets had a greater tendency to reduce their wealth by spending them. Over and above this influence, wealth varied by only about one-third of the amount that would have been expected based on forecasts of future income alone.

This implies that people who experienced sharp increases in income between 1995 and 2000 tended to borrow less than could have been justified with hindsight, unless they were already wealthy.

Comparing the computed change in wealth with the change in wealth that was actually observed shows these effects from a different perspective.

  • On average, household wealth increased by 1.5 times the average of incomes in 1999 and 2000; an increase of only 0.9 times would have been consistent with prudent spending and saving.
  • The average household in poverty in 1995 could, consistent with prudent spending and saving, have reduced its wealth by half of its average income in 1999 and 2000. In fact wealth rose by 1.4 times average 1999-2000 income.
  • Even households whose net debts exceeded their assets borrowed considerably less than could have been justified by their income growth between 1995 and 2000.
  • These calculations assumed that credit was available at reasonable interest rates of 7.5 per cent per annum on a long-term basis. Clearly the private sector does not lend to people in poverty on that basis and even rates charged by credit unions are considerably higher; the government may nevertheless be able to do so because its costs of collecting debt repayments are much lower and it is likely that the provision of extra credit to meet the needs of people in poverty would have to come from this source.


These calculations assumed that people in poverty borrowed on the basis of plausible forecasts about their future incomes. Some of those expectations were, however, bound to be disappointed: nearly a quarter of households in poverty in 1995 would, as a result of borrowing on the basis of the income projections generated, have accrued additional debts of more than 1.5 times the average of 1999-2000 income by 2000. These households were predominantly those who remained poor for the period 1995-2000. In 1995 their average income was £6,158, compared with £6,758 for the average household in poverty. By 2000 it had risen to £8,471, compared with £11,205 for the average household which had been in poverty in 1995.

Even then it is not clear that such debt levels are worse than the alternative of not being able to borrow. The average income of these households rose sharply in 2001 and 2002 so that by 2002 it was £12,891, compared with £13,412 for all households which had been in poverty in 1995. Households whose incomes take a long time to recover from low levels run up larger debts than those whose incomes recover rapidly. But the experience of these households does not demonstrate that above-average debt multiples are inherently unsupportable.

The broader perspective is that most households in poverty do not remain in poverty for very long. Figure 3 shows the proportion of the sample below the 1995 poverty threshold classified by the number of years between 1991 and 2002 in which they fell below the threshold. It can be seen that, for most working-age households, poverty was a temporary experience.

Figure 3: Indicators of persistence of poverty between 1991 and 2002

Conclusions and policy implications

The researchers conclude that the welfare of people in poverty would be improved if they had better access to credit on reasonable terms. The Social Fund exists to provide interest-free loans but the amounts involved are small compared with the figures identified here. The time over which credit is repaid is also short.

One policy option would be to extend greatly the scope of the Social Fund and to make repayments depend on people’s incomes while at the same time offering help with budgeting and money management. The advantage of this is that it would make extra credit available at little cost to the taxpayer; the disadvantage is that the collection of repayments on a basis related to income levels would appear as an extra form of means-testing. An appropriate balance between the two would need to be struck.

About the project

The researchers were James Mitchell, Kostas Mouratidis and Martin Weale of the National Institute of Economic and Social Research. The data were drawn from the British Household Panel Survey and relate to 987 records of households which provided complete data on income and wealth between 1991 and 2002 and whose heads were aged under 65 in 2002.